纯干货: kubernetes 多Master 二进制方式v1.20.11升级到v1.21.7

纯干货: kubernetes 多Master 二进制方式v1.20.11升级到v1.21.7

升级介绍

针对kubernetes小版本升级，只要更新相关二进制文件就可以了，由于kubernetes版本更新比较快，很多依赖还没解决晚上，所以不建议生产环境使用比较新的版本，升级目标小版本号不建议大于2。

kubernetes版本升级需要考虑到集群当中正在运行的容器收到影响，所以应该对各Node逐个进行驱离和驱逐Node上面的Pod。先更新Master服务，在逐个更新Node上的kubelet和kube-proxy服务。

升级规划

说明:

calico和etcd一般不需要更新。

升级准备

下载地址

wget https://dl.k8s.io/v1.21.7/kubernetes-server-linux-amd64.tar.gz

目前集群版本信息

# kubectl get nodeNAME STATUS ROLES AGE VERSIONtest-k8s-master-227 Ready master 18d v1.20.11test-k8s-master-230 Ready master 17d v1.20.11test-k8s-master-231 Ready master 17d v1.20.11test-k8s-node-228 Ready 18d v1.20.11test-k8s-node-229 Ready 18d v1.20.11test-k8s-node-233 Ready 16d v1.20.11# kubectl versionClient Version: version.Info{Major:"1", Minor:"20", GitVersion:"v1.20.11", GitCommit:"27522a29febbcc4badac257763044d0d90c11abd", GitTreeState:"clean", BuildDate:"2021-09-15T19:21:44Z", GoVersion:"go1.15.15", Compiler:"gc", Platform:"linux/amd64"}Server Version: version.Info{Major:"1", Minor:"20", GitVersion:"v1.20.11", GitCommit:"27522a29febbcc4badac257763044d0d90c11abd", GitTreeState:"clean", BuildDate:"2021-09-15T19:16:25Z", GoVersion:"go1.15.15", Compiler:"gc", Platform:"linux/amd64"}

开始升级

Master升级(在所有Master节点操作)

备份

mkdir /data/backup/kubernetes -pcp /data/local/kubernetes/bin/{kube-apiserver,kube-controller-manager,kube-scheduler} /data/backup/kubernetes/

停止master（先停止其中一台master,先升级一台，跑个测试服务，观察下是否正常，在升级其他master）

systemctl stop keepalived #可以先停掉keepalived，切走高可用VIP地址systemctl stop kube-apiserversystemctl stop kube-schedulersystemctl stop kube-controller-manager

升级master

tar xf kubernetes-server-linux-amd64.tar.gz/bin/cp kubernetes/server/bin/{kube-apiserver,kube-controller-manager,kube-scheduler} /data/local/kubernetes/bin/

启动keepalive和apiserver，其他组件

$systemctl start kube-apiserver$systemctl start kube-controller-manager && systemctl start kube-scheduler$kubectl get csWarning: v1 ComponentStatus is deprecated in v1.19+NAME STATUS MESSAGE ERRORcontroller-manager Healthy ok scheduler Healthy ok etcd-1 Healthy {"health":"true"} etcd-2 Healthy {"health":"true"} etcd-0 Healthy {"health":"true"} $systemctl start keepalived$kubectl version Client Version: version.Info{Major:"1", Minor:"20", GitVersion:"v1.20.11", GitCommit:"27522a29febbcc4badac257763044d0d90c11abd", GitTreeState:"clean", BuildDate:"2021-09-15T19:21:44Z", GoVersion:"go1.15.15", Compiler:"gc", Platform:"linux/amd64"}Server Version: version.Info{Major:"1", Minor:"21", GitVersion:"v1.21.7", GitCommit:"1f86634ff08f37e54e8bfcd86bc90b61c98f84d4", GitTreeState:"clean", BuildDate:"2021-11-17T14:35:38Z", GoVersion:"go1.16.10", Compiler:"gc", Platform:"linux/amd64"}

可以看到服务端组件已经升级成功。

升级另外两台Master

将新版本的报传到230和231

scp kubernetes-server-linux-amd64.tar.gz 192.168.253.230:/data/tools/scp kubernetes-server-linux-amd64.tar.gz 192.168.253.231:/data/tools/

停止230master （231一样的操作)

systemctl stop keepalived #可以先停掉keepalived，切走高可用VIP地址systemctl stop kube-apiserversystemctl stop kube-schedulersystemctl stop kube-controller-manager

升级

$cd /data/tools $tar xf kubernetes-server-linux-amd64.tar.gz $/bin/cp kubernetes/server/bin/{kube-apiserver,kube-controller-manager,kube-scheduler} /data/local/kubernetes/bin/ $systemctl start kube-apiserver $systemctl start kube-controller-manager && systemctl start kube-scheduler $kubectl versionClient Version: version.Info{Major:"1", Minor:"20", GitVersion:"v1.20.11", GitCommit:"27522a29febbcc4badac257763044d0d90c11abd", GitTreeState:"clean", BuildDate:"2021-09-15T19:21:44Z", GoVersion:"go1.15.15", Compiler:"gc", Platform:"linux/amd64"}Server Version: version.Info{Major:"1", Minor:"21", GitVersion:"v1.21.7", GitCommit:"1f86634ff08f37e54e8bfcd86bc90b61c98f84d4", GitTreeState:"clean", BuildDate:"2021-11-17T14:35:38Z", GoVersion:"go1.16.10", Compiler:"gc", Platform:"linux/amd64"}

Node组件升级

备份(所有node节点，包括master)

mkdir -p /data/backup/kubernetes/cp /data/local/kubernetes/bin/{kube-proxy,kubelet} /data/backup/kubernetes/cp /usr/bin/kubectl /data/backup/kubernetes/

隔离Node，让node变成不可调度的状态

kubectl cordon test-k8s-master-227kubectl get node NAME STATUS ROLES AGE VERSIONtest-k8s-master-227 Ready,SchedulingDisabled master 18d v1.20.11test-k8s-master-230 Ready master 17d v1.20.11test-k8s-master-231 Ready master 17d v1.20.11test-k8s-node-228 Ready 18d v1.20.11test-k8s-node-229 Ready 18d v1.20.11test-k8s-node-233 Ready 17d v1.20.11

停止(一台一台升级)

systemctl stop kubeletsystemctl stop kube-proxy

升级

cd /data/tools/tar xf kubernetes-server-linux-amd64.tar.gz/bin/cp kubernetes/server/bin/{kubelet,kube-proxy} /data/local/kubernetes/bin//bin/cp kubernetes/server/bin/kubectl /usr/bin/ #升级node 可以不用拷贝kubectl

启动

systemctl daemon-reload && systemctl start kubeletsystemctl start kube-proxysystemctl status kube-proxy● kube-proxy.service - Kubernetes Proxy Loaded: loaded (/usr/lib/systemd/system/kube-proxy.service; enabled; vendor preset: disabled) Active: active (running) since Mon 2021-12-06 14:12:07 CST; 7s ago Main PID: 85281 (kube-proxy)

启动成功，版本也改过来了

kubectl get nodeNAME STATUS ROLES AGE VERSIONtest-k8s-master-227 Ready,SchedulingDisabled master 18d v1.21.7test-k8s-master-230 Ready master 17d v1.20.11test-k8s-master-231 Ready master 17d v1.20.11test-k8s-node-228 Ready 18d v1.20.11test-k8s-node-229 Ready 18d v1.20.11test-k8s-node-233 Ready 17d v1.20.11

解除隔离

kubectl uncordon test-k8s-master-227

其他节点一样的操作

将包分发到所有节点，其他步骤参考上面走一遍

scp -P 22 kubernetes-server-linux-amd64.tar.gz 192.168.253.228:/data/tools/scp -P 22 kubernetes-server-linux-amd64.tar.gz 192.168.253.229:/data/tools/scp -P 22 kubernetes-server-linux-amd64.tar.gz 192.168.253.230:/data/tools/scp -P 22 kubernetes-server-linux-amd64.tar.gz 192.168.253.231:/data/tools/scp -P 22 kubernetes-server-linux-amd64.tar.gz 192.168.253.233:/data/tools/

一番猛如虎的操作，所有node均已升级完成

kubectl get nodeNAME STATUS ROLES AGE VERSIONtest-k8s-master-227 Ready master 18d v1.21.7test-k8s-master-230 Ready master 17d v1.21.7test-k8s-master-231 Ready master 17d v1.21.7test-k8s-node-228 Ready 18d v1.21.7test-k8s-node-229 Ready 18d v1.21.7test-k8s-node-233 Ready 17d v1.21.7

测试

$vim test.yaml apiVersion: v1kind: Podmetadata: name: busybox namespace: defaultspec: containers: - name: busybox image: busybox:1.28.3 command: - sleep - "3600" imagePullPolicy: IfNotPresent restartPolicy: Always $kubectl apply -f test.yaml $kubectl get po -o wideNAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATESbusybox 1/1 Running 0 107s 10.244.4.3 test-k8s-node-233 nginx 1/1 Running 0 28m 10.244.39.132 test-k8s-node-229

pod 也可以正常启动

至此，二进制版本升级完毕

更多文章请扫一扫